Combing for a signal buried in noise

Picking out a signal from a sea of noise is a ubiquitous research challenge. If the signal recurs, multiple measurements can be appropriately averaged to improve the signal-to-noise ratio. For a single transient event, Fourier decomposition—breaking the signal down into its frequency components—can help isolate signal from noise. The signal’s frequency components are correlated in phase with each other, whereas those of the noise are not. When the components are summed as complex numbers that encode phase and amplitude, only the signal adds up coherently. Calculating the Fourier transform works well enough for slow signals lasting a few microseconds, but faster signals run up against the resolution limits of both detectors and analog-to-digital converters. Now a University of California, San Diego, group led by Stojan Radic has employed tunable optical frequency combs, developed in Radic’s lab, to catch those faster signals. (For more on frequency combs, see Physics Today, June 2000, page 19 .) The researchers used two phase-locked combs with slightly different spacing between their teeth. They combined a test signal—an 80-ps microwave pulse laden with white noise—with the first comb to make copies of the signal at each tooth. Using the second comb’s teeth as reference frequencies, the researchers measured a narrow slice of each copy and thus partitioned the data into frequency bins. The outputs of those bins could then be summed. The gain in signal-to-noise ratio is limited only by the number of copies one can generate. The researchers have tested up to 300 copies for a gain of 25 dB. The group’s tunable frequency combs are already commercially available, and Radic is eager to see how other researchers put his setup to use. (V. Ataie et al., Science 350, 1343, 2015, doi:10.1126/science.aac8446 .)